Author Topic: Bench PSU kit comments?  (Read 17881 times)

0 Members and 1 Guest are viewing this topic.

Offline arnoldus

  • Newbie
  • Posts: 1
Bench PSU kit comments?
« on: June 13, 2010, 06:45:25 pm »
I stumbled across this kit (30V-2A):
http://tuxgraphics.org/electronics/201005/bench-power-supply-v3.shtml#3lfindex9

It's quite cheap, even when counting the costs of the additional needed parts.

I wanted to know 1) if you think there are better readybuilt or kit alternatives for this price
2) If the method of using pulse modulation to generate the voltages is a proper way? Does it have drawbacks?
 

alm

  • Guest
Re: Bench PSU kit comments?
« Reply #1 on: June 13, 2010, 07:30:05 pm »
I dislike this method of control, it looks almost the same as his previous version. My main issue with this design is that the ADC of the AVR is pretty slow, max 10kS/s or so. Plus the base current will only be increased/decreased by one DAC unit (of the ~3600 units between fully on and fully off). This means that it takes about .4s to go from no base current to fully saturated. It will probably have a hard time tracking even simple 50/60Hz ripple! There is (at least in the previous version) a separate regulation that will shut it down in case of a dead short, but in any other case it's really slow. Even a crappy opamp like the ua741 will easily have thousand times the slew rate. In my opinion the right way to design a programmable power supply is to use DAC's to generate a reference voltage, but have op-amps do the actual feedback loop. Op-amps are much better at this type of work than a micro controller, this seems just doing it digitally for the sake of digital, almost as bad as using an Arduino to blink a LED ;). Plus I dislike having safety features like current limits implemented in software (cf. Therac-25). Something like an LM723, L200 or double LM317 will have a much better regulation.
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 30100
  • Country: au
    • EEVblog
Re: Bench PSU kit comments?
« Reply #2 on: June 13, 2010, 07:39:44 pm »
I dislike this method of control, it looks almost the same as his previous version. My main issue with this design is that the ADC of the AVR is pretty slow, max 10kS/s or so. Plus the base current will only be increased/decreased by one DAC unit (of the ~3600 units between fully on and fully off). This means that it takes about .4s to go from no base current to fully saturated. It will probably have a hard time tracking even simple 50/60Hz ripple! There is (at least in the previous version) a separate regulation that will shut it down in case of a dead short, but in any other case it's really slow. Even a crappy opamp like the ua741 will easily have thousand times the slew rate. In my opinion the right way to design a programmable power supply is to use DAC's to generate a reference voltage, but have op-amps do the actual feedback loop. Op-amps are much better at this type of work than a micro controller, this seems just doing it digitally for the sake of digital, almost as bad as using an Arduino to blink a LED ;)

I agree. Much better to do it with opamps. I'm working on just such a neat PSU design at the moment actually.

Dave.
 

Offline Anders

  • Contributor
  • Posts: 34
  • Country: se
Re: Bench PSU kit comments?
« Reply #3 on: June 15, 2010, 09:41:43 am »
When do we get to see this PSU Dave?  :)

I also agree on the OP-Amp solution even though you (if you must use a micro) could choose a PIC which has faster ADC than the AVR.
Proper Planning Prevents Piss Poor Performance
 

alm

  • Guest
Re: Bench PSU kit comments?
« Reply #4 on: June 15, 2010, 02:45:10 pm »
Sure, and a less brain-dead control algorithm could also improve things. But that's entirely missing the point. The point is that it's stupid to implement an analog feature with dedicated IC's for a few cents available in software. It's not like there's some complex DSP taking place, it's just 'if input > threshold + hysteresis than dac_output += 1; elseif input < threshold - hysteresis than dac_output -= 1'. That's exactly what an op-amp does, and even the really crappy ones will have a gain-bandwidth product easily into the hundreds of kilohertz.
 

Offline jimmc

  • Frequent Contributor
  • **
  • Posts: 278
  • Country: gb
Re: Bench PSU kit comments?
« Reply #5 on: June 17, 2010, 06:43:55 pm »
Interesting chip here http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1003,C1040,C1055,P38503

Makes for a simple PSU without the 1.2v minimum of the LM317 etc.

Jim
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 30100
  • Country: au
    • EEVblog
Re: Bench PSU kit comments?
« Reply #6 on: June 18, 2010, 01:09:46 am »
When do we get to see this PSU Dave?  :)

Hopefully in the not too distant future.

Dave.
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 13177
  • Country: gb
  • 0999
Re: Bench PSU kit comments?
« Reply #7 on: June 18, 2010, 08:05:55 am »
Interesting chip here http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1003,C1040,C1055,P38503

Makes for a simple PSU without the 1.2v minimum of the LM317 etc.

Jim

Sorry that's not suitable for this application. It has a maximum input voltage of only 36V, a maximum current of only 1.1A and requires a minimum load of 0.5mA to regulate properly.

 

alm

  • Guest
Re: Bench PSU kit comments?
« Reply #8 on: June 18, 2010, 11:13:10 am »
What application? Arnoldus asked for opinions on a kit that offered 30V/2A, so 36V input voltage would be plenty, although 1A max is a bit low (but you can parallel them). The main issue for a bench supply is that there doesn't appear to be an easy way to add a programmable current limit. The 0.5mA minimum is easily satisfied by including a load resistor or a power LED.
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 13177
  • Country: gb
  • 0999
Re: Bench PSU kit comments?
« Reply #9 on: June 18, 2010, 02:27:04 pm »
What application? Arnoldus asked for opinions on a kit that offered 30V/2A, so 36V input voltage would be plenty, although 1A max is a bit low (but you can parallel them).
Sorry, you're right, I was confused with another thread.

Quote
The 0.5mA minimum is easily satisfied by including a load resistor or a power LED.
It won't be possible to ensure a minimum load current with just resistor and an LED because the current will fall as the output voltage is lowered and will be zero when the voltage is below the LED's forward voltage.

It's possible to go down to 0V using an LM317 but another op-amp is required.
http://www.silicontronics.com/index.php?action=ezportal;sa=page;p=19
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 13177
  • Country: gb
  • 0999
Re: Bench PSU kit comments?
« Reply #10 on: June 23, 2010, 12:22:21 pm »
You need a constant current sink to go down to 0V which unfortunately requires a negative supply.
 

alm

  • Guest
Re: Bench PSU kit comments?
« Reply #11 on: June 23, 2010, 12:53:29 pm »
That same tricks works with any old three-terminal regulator like the LM317, see the OnSemi LM317 datasheet. So the down to 0V is just a lame marketing claim, without negative current, it will only go down to 1.2V like most linear regulators.
 

Offline slburris

  • Frequent Contributor
  • **
  • Posts: 522
Re: Bench PSU kit comments?
« Reply #12 on: June 23, 2010, 02:59:33 pm »
Seriously, this design inserts a microcontroller in the feedback loop of
the voltage regulation?  That's fine if you have a constant load, like
maybe a light bulb (after it warms up), but most real loads change
over time. 

Putting a micro in the control loop means it will take some number
of milliseconds to respond to a load change.  That's a terrible transient
response.  And what's with the statement:

"Fast means within milliseconds (as fast as an operational amplifier)."

Eh?  I'm not much of an analog guy, but opamps are much faster.

This design implements a DAC with both PWN and an R2R ladder?  And
oversamples the ADC to get more resolution?  In my opinion, resorting
to stuff like this means one of two things.

1)  You are stuck with an existing microcontroller and have to shoehorn
theses functions in.

2)  You are a fan of a particular micro and that blinds you to the fact
that you've chosen the wrong chip!  Didn't Dave rant about this in an
early episode about PIC vs AVR?  You should pick a chip with the
appropriate ADC and DAC functions, or implement them in some
other way rather than trying to hack a solution together.

You can get a 12 bit serial DAC from microchip for about $1.  For $3
you can upgrade to 16 bits.  Seems like an easy choice to me.

Going back to the control loop, why not just pick an adjustable
linear regulator and couple the DAC output to the adjust pin of
the regulator, and let the regulator deal with the transient response issues,
short circuit protection, over temp protection, etc.  These are all
well solved problems and pushing them into software hurts
reliability.  What happens if you have a bug and the micro stops
responding?

PWM is fine for motor control but it's not such a good choice to generate
precise voltages.  As the site points out, there are filtering and response
issues.  Well, it's pretty easy to build 20bit or better DACs not with PWM,
but with a delta-sigma modulator.  The filtering problems are simpler, see
http://jjmk.dk/MMMI/Exercises/05_Counters_Shreg/No7_PWM_vs_SigmaDelta/index.htm
for a quick overview.  Of course there's no hardware support for this
in the AVR, so it would probably be too slow.

Again, just pick the right chips for what you need.

How about this?  The April 2010 issue of Elektor describes the UniLab,
a *dual* 0-30v 3amp power supply:

http://www.elektor.com/magazines/2010/april/unilab.1285365.lynkx?tab=1

It uses a National Semiconductor simple switcher chip (I've used those --
they are very handy).  No huge heat sinks to deal with, very efficient, although
it's going to be noisier than a linear supply.  And it goes down to zero volts.

Of course if you pay Elektor's prices for the complete kit of parts, you could just
buy a Mastech power supply, like HY3003D for a nice linear supply with nice
displays.

http://www.mastechpowersupply.com/dc-power-supply/linear-power-supply/mastech-regulated-variable-dc-power-supply-hy3003d-30v-3a/prod_2.html


Scott

 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 13177
  • Country: gb
  • 0999
Re: Bench PSU kit comments?
« Reply #13 on: June 24, 2010, 08:52:40 pm »
Hmm, aren't Mastech cheap and cheerful Chinese junk? I have a Mastech meter and whilst it's good for taking rough measurements, I wouldn't trust it too much. I'd rather build something I've designed so I can ensure it's good.

That same tricks works with any old three-terminal regulator like the LM317, see the OnSemi LM317 datasheet. So the down to 0V is just a lame marketing claim, without negative current, it will only go down to 1.2V like most linear regulators.

A part from one subtle difference, the circuit on the LM317 uses a voltage reference diode, the circuit I posted previously uses a constant current sink to ensure a minimum load for a regulator which can already do 0V. The circuit on the LM317 is inferior because of the component tolerances: if the regulator is 1.3V and the LM117 reference is 1.2V the minimum voltage will be 100mV. Using an op-amp to shift LM317's reference below 0V is a better option but it requires many extra parts.
 

Offline slburris

  • Frequent Contributor
  • **
  • Posts: 522
Re: Bench PSU kit comments?
« Reply #14 on: June 24, 2010, 09:19:02 pm »
I asked for and received the schematic for the HY3005F-3 power supply.

The 5V supply is a LM723 with a 2N3055 pass transistor, pretty conventional.
All the digital displays are driven by 7107s.  The variable supplies are 741 opamps
driving 3 parallel 2n3055's with load sharing emitter resistors, nothing wacky or
unexpected about that.

So at least on a schematic basis, this power supply doesn't look too dodgy,
but since I haven't seen one in person, don't know about the build quality.

Scott

Hmm, aren't Mastech cheap and cheerful Chinese junk? I have a Mastech meter and whilst it's good for taking rough measurements, I wouldn't trust it too much. I'd rather build something I've designed so I can ensure it's good.


 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: Bench PSU kit comments?
« Reply #15 on: June 24, 2010, 11:44:03 pm »
I purchased one, and should be able to examine the guts to describe it, in 2 weeks or so.  But various older models have been dissected in other forums such as RC group.  If you feel the regulator design is inadequate its design is simple and easy enough to modify.

If you total the costs of buying parts to build: dual LED DVM, transformer, casing, knobs, PCBs etc., I doubt you can match price nor obtain a polished look by buying a Mastech and modifying it as you see fit.

Although I cannot be dead certain, there are so many other PSU's that look, and spec-a-like the Mastech's to suggest the principle ODM provides PSU to other brand names, such as the low end Instek models, that suggests the fundamentals of Mastech's product are sound.

A $90 Mastech Model GDS-3030D versus an identical Model numbered Instek for $180 with similar power supply specs, but with differences such as handle, the single LED readout and in the manual, there are addition controls in the back of the Instek absent in the Mastech.



http://www.rcgroups.com/forums/showthread.php?t=849073

A view of the main circuit board of one model





I asked for and received the schematic for the HY3005F-3 power supply.

The 5V supply is a LM723 with a 2N3055 pass transistor, pretty conventional.
All the digital displays are driven by 7107s.  The variable supplies are 741 opamps
driving 3 parallel 2n3055's with load sharing emitter resistors, nothing wacky or
unexpected about that.

So at least on a schematic basis, this power supply doesn't look too dodgy,
but since I haven't seen one in person, don't know about the build quality.

Scott

Hmm, aren't Mastech cheap and cheerful Chinese junk? I have a Mastech meter and whilst it's good for taking rough measurements, I wouldn't trust it too much. I'd rather build something I've designed so I can ensure it's good.



« Last Edit: June 24, 2010, 11:56:02 pm by saturation »
Best Wishes,

 Saturation
 

Offline slburris

  • Frequent Contributor
  • **
  • Posts: 522
Re: Bench PSU kit comments?
« Reply #16 on: June 25, 2010, 03:47:19 am »
It sounds like the rcgroups folks are having problems with their power supplies because they
are backfeeding power from a battery into the power supply, reversing polarity, etc.

Not my use case, I just want to power circuits for testing.

At least for the analog Mastech supplies I've seen schematics for, there don't
appear to be any hard to obtain parts (well maybe the transformers, but I can't
think of any transformer ever going bad), so I've got pretty high confidence I could
repair them, even as analog challenged as I am :-)

I like the cable dressing in the pics you attached.  That shows attention to
detail, not something I would expect if Mastech is at the low end of the power supply
foodchain.  Why are they perceived as junk?   

When you get yours, let us know what you think.  Thanks.

Scott
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: Bench PSU kit comments?
« Reply #17 on: June 25, 2010, 10:43:23 am »
Yes, but the good news is that these folks are abusing their units and it survives, and if repairs are needed, the genericness you describe of the design makes it fairly easy.

I'm not sure if Mastech has a reputation for making junky stuff, as one poster here suggests, its one reason I'm posting what other folks have already done to them and a look side.

The RCgroup has many threads, apparently they like Mastech for charging their batteries, and it goes back to 2004.  They've established a good working relationship with the RC folks since then:

http://www.rcgroups.com/forums/showthread.php?t=330215

http://www.rcgroups.com/forums/showpost.php?p=1748079&postcount=2

As for some non user error PSU failures you read there, know between 2000-7 was the era of the capacitor plague, and quite a number of devices blew power caps in their power supplies beyond Mastech, commonly PC power supplies, then, DVD players, etc., anything not powered by a wall wart was above normal expectations.

http://en.wikipedia.org/wiki/Capacitor_plague

Worry not about transformers, you can get them anywhere, the problem is the cost.  A typical laminated core 2.8A is about $20, but a 3A is about$40.  Make the same torroidal core and nearly 1.5-x3 the price. The case alone is about $15-20. Just adding those parts costs together and considering a 3A Mastech costing $50-100 assembled, building from scratch is not a cost saving idea, but modding it to improve its performance is a more cost effective.

A one stop solution is Digikey.com but you'd want to shop around for better prices.

Although being able to swing from zero to hero volts is a useful feature, its less so than keeping ripple down, and to have tight load and line regulation.  Few designs work with less than 3V or so, and when they do, they often use batteries and have a DC-DC conversion stage to bring circuit voltages up.  So to emulate a battery, you don't want ripple.  NiMH drain should cutoff at 0.9V, a design intending to down to 1.0V, should, for the sake of the battery, shut down at 0.9V to prevent damage to the NiMH.















It sounds like the rcgroups folks are having problems with their power supplies because they
are backfeeding power from a battery into the power supply
, reversing polarity, etc.

Not my use case, I just want to power circuits for testing.

At least for the analog Mastech supplies I've seen schematics for, there don't
appear to be any hard to obtain parts (well maybe the transformers, but I can't
think of any transformer ever going bad)
, so I've got pretty high confidence I could
repair them, even as analog challenged as I am :-)

I like the cable dressing in the pics you attached.  That shows attention to
detail, not something I would expect if Mastech is at the low end of the power supply
foodchain.  Why are they perceived as junk?    

When you get yours, let us know what you think.  Thanks.

Scott

« Last Edit: June 25, 2010, 11:33:22 am by saturation »
Best Wishes,

 Saturation
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: Bench PSU kit comments?
« Reply #18 on: July 02, 2010, 06:40:16 pm »
I finally received the unit pictured below and here's initial impressions.

Buying from Mastech was a good experience, but email that unit is in stock before purchasing as the website inventory count maybe inaccurate. Sean is very responsive to email; you'll get an answer < 1 hour.  We've had several exchanges before payment.  The website has some errs, but the picture of the device below is true.

It shipped via Fedex, from CA.  It took a week to arrive to me.

It was a sealed box original from China.  It was packed well, and interestingly the styrofoam mold was hand cut and assembled with glue to fit the unit.

I haven't had time to photograph anything, but build quality and design is good for any design, but excellent given the cost. 

The unit is about the size of a shoebox on its side.

The design is like an old IBM PC case, all sheet metal held down by screws.  5/6 sides are all metal walls, the front face is plastic attached to a metal frame.  The machining, fit, knobs, buttons, jacks, paint and finish are all well made, like a good Heathkit circa 1980s, very solid feel.  Plastic is thick and hefty, a dull off white finish.

Dials turn nicely, panel meters are evenly lit.

The manual is chinglish and not helpful.  None of its advertised features, and more, are mentioned.   Using it is fairly straightforward, and I am putting together what it does by reading other Mastech and the GWInstek manual.  For example, it does not describe what the hi/lo amps switch does, the adjustment screw beneath the volt or ammeters, whether it had short circuit protection, or how to remove the ingeniously placed fuse holder. 

The fan is strong, makes for noisy fan like sound.  You can feel the suction from the vents cut into the sides of the unit, making for palpable airflow, exhausting from the rear.  Older Mastech designs had a large heat sink in the rear which made the unit bulkier, so the fan makes the unit much more compact.  The noisy isn't any more than a PC would make.

The output voltage/current is fairly stable after 3 hours of testing, no drift.  The meters are at least accurate to 0.3%, but I haven't completed more thorough electronic testing, and follows later.  Fine adjusts about 10% of coarse, so at 30 fine is 3, at 3 fine is 0.3 etc.,

Hi amps is 1.5-3 A, Lo is 0- 1.5 amps.

It will work down to 0 volts.

It has short circuit protection.

Constant voltage and current lights work when the modes engage.

Opening the case is easy.  Sheet metal screws. Rubber feet screwed to bottom.  The PC board is built like the photo from RC groups, clean, widely spaced, through hole parts typical of 1980s style design.  Easy to fix if needed.  My prelim scan confirms many easy to find generic parts.  Well assembled, clean, good quality hand soldering.  Wires cleaned routed and bunched.  Nothing unsafe needed fixing.

Main transformer is laminated core.  Full wave bridge is screwed into the metal floor to heat sink it and the leads are insulated with shrink wrap tubing.  Main power transistors are bolted to an aluminum heatsink the size of the entire rear, and faces the fan. 

In toto, very well made linear PSU.  As mentioned earlier, well worth buying and modding as you see fit.  You may not be able to get a replacement transformer if it blows, but its easy to fit a substitute in here.  The overall look is very professional.

To follow: load regulation, drift under load tests, and more on DVM accuracy.















I purchased one, and should be able to examine the guts to describe it, in 2 weeks or so.  But various older models have been dissected in other forums such as RC group.  If you feel the regulator design is inadequate its design is simple and easy enough to modify.

If you total the costs of buying parts to build: dual LED DVM, transformer, casing, knobs, PCBs etc., I doubt you can match price nor obtain a polished look by buying a Mastech and modifying it as you see fit.

Although I cannot be dead certain, there are so many other PSU's that look, and spec-a-like the Mastech's to suggest the principle ODM provides PSU to other brand names, such as the low end Instek models, that suggests the fundamentals of Mastech's product are sound.

A $90 Mastech Model GDS-3030D versus an identical Model numbered Instek for $180 with similar power supply specs, but with differences such as handle, the single LED readout and in the manual, there are addition controls in the back of the Instek absent in the Mastech.




[/quote]
Best Wishes,

 Saturation
 

alm

  • Guest
Re: Bench PSU kit comments?
« Reply #19 on: July 02, 2010, 07:23:06 pm »
Thanks for posting your impressions! Does the Hi/Lo halve the max voltage and put the windings in parallel instead of series? Otherwise I don't understand the point (except doubling the range of the current pot). Sounds a bit awkward if you're just around the 1.5A point, I would have designed some overlap if I had to do it at all (and preferably switch automatically). I dislike fans in equipment, especially fixed speed ones, so I'm not sure if I'd be happy with the trade-off, but I'm sure it reduces the cost (fans are cheaper that large heat sinks).
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: Bench PSU kit comments?
« Reply #20 on: July 02, 2010, 08:45:44 pm »
Hi Alm,

My pleasure.  I enjoy finding a good deal. 

Alas, no, it doubles the pots capability, which is strange.  By putting the secondaries in parallel they could double the output current while halving the maximum voltage but!  ???  The fan is very generic, its an easily obtainable PC case fan, I have the exact same model I use to supplement my PCs cooling, cost me $3 for 2.  There is enough room inside the case to add mods.  You could add a temperature driven variable speed control, to cut noise down.

The quiescent power is 15 watts, measured with a power meter.

This is what the older design looked like, and heat sink does bulk up the design.  Spin the flash graphic for a 360 deg view.

http://www.mpja.com/email/flash/showflash.asp?itemno=14601%20PS



Thanks for posting your impressions! Does the Hi/Lo halve the max voltage and put the windings in parallel instead of series? Otherwise I don't understand the point (except doubling the range of the current pot). Sounds a bit awkward if you're just around the 1.5A point, I would have designed some overlap if I had to do it at all (and preferably switch automatically). I dislike fans in equipment, especially fixed speed ones, so I'm not sure if I'd be happy with the trade-off, but I'm sure it reduces the cost (fans are cheaper that large heat sinks).
Best Wishes,

 Saturation
 

Offline johnnyk126

  • Contributor
  • Posts: 9
Re: Bench PSU kit comments?
« Reply #21 on: October 01, 2010, 04:38:40 pm »
When do we get to see this PSU Dave?  :)
Hopefully in the not too distant future.
How is that project going? Can you give some more information when and in what form would it be available?
And possibly a little bit more about the design...
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf